Update Dockerfile

Dockerfile

@@ -2,13 +2,14 @@ FROM python:3.10-slim
 
 WORKDIR /app
 
-# System dependencies
+# System dependencies (including image codecs)
 RUN apt-get update && apt-get install -y \
     libgl1 \
-    libglx0 \
     libglib2.0-0 \
     libgomp1 \
     libopenblas0 \
+    libjpeg62-turbo \
+    zlib1g \
     && rm -rf /var/lib/apt/lists/*
 
 # Create writable directories
@@ -25,9 +26,14 @@ RUN pip install --no-cache-dir --upgrade pip setuptools wheel && \
         --extra-index-url https://download.pytorch.org/whl/cpu && \
     pip install --no-cache-dir -r requirements.txt
 
+# Pre-download Qwen2-VL model (avoids first-request timeout)
+RUN python -c "from transformers import AutoModelForCausalLM, AutoProcessor; \
+    AutoModelForCausalLM.from_pretrained('Qwen/Qwen2-VL-2B-Instruct', device_map='cpu'); \
+    AutoProcessor.from_pretrained('Qwen/Qwen2-VL-2B-Instruct')" || echo "Pre-download skipped"
+
 COPY . .
 
-# ── Environment Variables ─────────────────────────────────────────────────────
+# Environment Variables
 ENV PYTHONUNBUFFERED=1
 ENV PORT=7860
 ENV TRANSFORMERS_CACHE=/.cache
@@ -36,24 +42,24 @@ ENV EASYOCR_CACHE=/.cache
 ENV FEEDBACK_DATA_PATH=/app/data
 ENV PADDLE_DOWNLOAD_CACHE=/.cache/paddlepaddle
 
-# CPU Thread Optimization (Crucial for 2-vCPU)
+# CPU Thread Optimization
 ENV OMP_NUM_THREADS=1
 ENV MKL_NUM_THREADS=1
 ENV OPENBLAS_NUM_THREADS=1
 
-# ── VLM Optimization ──────────────────────
+# VLM Configuration
 ENV ENABLE_VLM=1
 ENV VLM_MODEL_ID=Qwen/Qwen2-VL-2B-Instruct
-ENV VLM_MAX_NEW_TOKENS=256 
-# UPDATED: Increased timeout to 150s to avoid fallback to messy OCR
+ENV VLM_MAX_NEW_TOKENS=256
 ENV VLM_TIMEOUT_SECONDS=150
-# Self-disable VLM if RSS exceeds 12 GB
 ENV VLM_MEMORY_LIMIT_MB=12000
 
+# Disable Paddle (saves memory)
+ENV ENABLE_PADDLE=0
+
 EXPOSE 7860
 
-# ── Startup command ───────────────────────────────────────────────────────────
-# Increased timeout-keep-alive for the 4GB weight download
+# Startup command
 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860", \
      "--workers", "1", \
      "--timeout-keep-alive", "300"]